Alternate Action Recommender System Using Recurrent Patterns of Smart Home Users

We present an alternate action recommender system for Internet of Things (IoT) smart home users. Our system takes the data of the IoT devices of the smart home users as input, applies our custom pattern-mining algorithm to derive the highly probable and active recurrent patterns of an individual user, and finally recommends the alternate possibilities of achieving the deviated actions from any of the active patterns. The derived active patterns are very specific to the users, which are based on their usage of the IoT devices in their smart home ecosystem, thus making the recommendations personalized. In our system, we also consider the context of the smart home environment through the IoT device state and criticality of the recommendations in the user's smart home system where these parameters play crucial role. We have designed and implemented our core algorithm by modifying the SMILE-ARM algorithm to better suit the IoT devices data of the smart home ecosystem and validated it using user trial methods and standard algorithm validation techniques.