|Objectives||Many IoT devices has been developed and data integration activities has been reported in AgWG. But there is no open big data that AgWG members can use in free. In this session, reports on observing massive data using IoT/WoT, creating big data using thematic technology, common APIs to ensure accessibility interoperability, comparability of data, AI applications using big data, etc. are expected.
|1. ||Multi-purpose Smart Irrigation Valve (SIV) - (PDF)|
The existing irrigation valves control system is a very complex model, it need external sources like, wireless sensor networking, soil moisture Sensor, evapotranspiration sensor, drones and satellites to operate. Since, it has too many wireless networking, which decreases its commercial viability and usability. In this work, we propose a Smart Irrigation Valve (SIV) that can avoid the usage of all external sources in controlling the valve reducing the overall system complexity. One SIV device can handle four normal valve operations supporting scalable deployment that can cover entire farmland. In addition to the regular irrigation operation, this device can help for crop like paddy to maintain the prescribed water height for the plants at each growth stage. Since it is an IoT (Internet of Things) enabled device, data such as acreage, crop cultivated, farm location etc., collected from the farmland can be shared via cloud and further, it can help as a platform for farmers to reach out potential buyers. The soil data can also be collected helping with the verification of insurance claims due to crop damage from floods, drought and similar weather conditions helping farmers in insurance policy coverage while providing the insurance provider with more information in settlements. The SIV approach presented uses the free satellite reflected signal from the soil to measure the soil moisture. Based on the soil moisture, the valve will be actuated automatically. A single device can cover 80% of an acre diameter, with a minimum of 1000 samples for every one hour. The prototype has been developed and tested for more than 4 months at 7 different farm environments.
|2. ||Research on Agriculture Knowledge Service System Design - (PDF)|
Based on introduction to agriculture data application in the world. The paper point out that agriculture data in Sichuan province which located in southwest of China is lack of planning, integration and organization. Then face to the Sichuan province agriculture data application requirement, the paper proposed Sichuan Province Agriculture Knowledge Service System’s framework. It divides into 3 layers, basic database layer, knowledge discovery layer and user layer.
Furthermore, discus 3 key problems: information organization, data indexing and linked open data in open data. Finally, propose the intending contents to be research in this system.
|3. ||Experimentation and the Diffusion of Technology in China Using Big Data to explore Consumer Channel Choice - (PDF)|
A key challenge for academic research on behavior in society is access to data of the required volume, variety and veracity. Such data is of quantifiable operational value to corporate data owners, however, unlocking its strategic value through inductive ‘Big Data’ analysis invites a company to give access to that data before the bounds on data value can be fully assessed. This requires high degrees of trust. In this paper we report results from a decade of infrastructure and trust building that has allowed exploration of technology diffusion in China, the world’s largest market for digital technologies. We innovate by including in our sample both early adoption and early rejection of a new technology to enable characterization of the first stage of the diffusion process: ‘Experiri’ – to try out. We explore this by examining consumer channel choice over a period of two years for over 1 million individual consumers and over 1000 products. Using a highly scalable data-mining tool we demonstrate that propensity to experiment with new channels is predictable and has a dependency on the demographics of the consumer. This offers the possibility of better management of new technology introduction and market development.
|4. ||Development of Data Cloud Management Platform for Forest Ecosystem Observation Station - (PDF)|
Ecosystem located observation is an effective way to continuously obtain forest ecological data. With the development of sensor technology, the amount of data is increasingly accumulated and the types of data observed become more and more. With the rapid accumulation of forest ecological observation data, a new mechanism is needed for data management and data sharing.
This talk developed a data cloud management platform, which can effectively manage the observation data from forest ecosystem located observation station network. The platform uses the Internet of things technology to realize data wireless transmission, uses cloud storage technology to realize data cloud storage, uses cloud computing technology to implement cloud analysis and visualization of observed data, and further to achieve a wide range of data sharing through data services、data downloads and other ways.
The system provides services for professionals in monitoring the condition forest ecosystem. The software functions include real-time monitoring of ecological location monitoring data, pooling of data collection, data query, visualization and data sharing, and extensive users, devices and system security settings.
|1. ||Deep learning based preliminary research on recognition of Golden Monkey sound|
Golden monkey is an endangered and rare animal protected by the State. It is extremely meaningful for the protection of golden monkeys to effectively monitor their activities through sound monitoring and recognition. In recent years, several Deep Learning models are used for recognition of image, speech and text, which has made a great breakthrough with extensive study of Deep Learning. Therefore, aiming at the study of golden monkey sound recognition, this paper presents a method of using open source Deep Learning model to build a stronger and more robust environmental sound recognition model. Meanwhile this paper discusses the influence of model accuracy when training sound recognition model on the number of different neuron nodes and the number of different training layers. The research provides effective technical support for monitoring and protection of golden monkeys.
|2. ||A Proof of Stake Sharding Protocol for Scalable Blockchains (replay) - (PDF)|
Cryptocurrencies such as Bitcoin has drawn great attention recently. The public ledger blockchain serves as a secure database for cryptocurrencies. However, only 3 to 7 transactions can be processed per second, which means the blockchain does not scale. To address this problem, we propose a new consensus protocol based on sharding and proof of stake. The scalability of our proposed method is expected to increase linearly with the network size. We discuss proposed method from the scalability evaluation, complexity and security view.
|3. ||Field Phenomics Research (Tentative)|
|4. ||Plan to add metadata to Field Server Data using RDF - (PDF)|
Field Server is a Web of Things for Agriculture developed in NARO and data is available at http://fsds.dc.affrc.go.jp/data[1-4]. Data is consisted with 40 million raw data files and 60 million camera files. Unfortunately, metadata not provided,. It is difficult to use Field Server data metadata not provided. To add metadata for Filed Server Data, I’m planning to add RDF files in each directory, referencing existing web ontology for tag names and point to other RDF. For example, where Field Sever is deployed should refer gaonames.org.