Data Pipeline for Generation and Recommendation of the IoT Rules Based on Open Text Data

Nowadays, many devices and sensors we use in everydaylife are connected to the internet. We call this the IoT (Internet ofThings). With more things being used, it is getting more difficultfor users to use them efficiently. Without overcoming thischallenge, IoT cannot be vitalized. To solve this problem, manyvoice agent systems, including Apple Siri and Amazon Alexa, areextending their service domains to cover the IoT environment. However, given the fact that IoT is in its infant stage, usersencounter a number of challenges when using it, as most agentsystems are designed to react to the user's utterance. In this paper, we propose a method to generate automatic rules with the user'sthings and made a suggestion on the best rule that is generated forthe user to use without having to make extra effort. To do this, webuilt an ontology model for the IoT things-action-context and adata pipeline for mining open Web data regarding rules - automatictasks, based on open data. After gathering open Web data, we usedthe natural language processing technology to make IoT rules bymapping the open Web data with our ontology model. With thisapproach, when a user buys new things and installs them in his/herenvironment, the user will get the recommendation from our systemproactively regarding the automatic tasks which include new thingscombined with the existing user things. As a result, a user becomesable to use his/her things very efficiently without having muchknowledge of the things. We believe that our proposed methodwould greatly rev up the use of IoT.