N-Gram based smart living machines (SLM) on IOT platform

Abstract

In our previous work, we have presented a pioneering step in designing Bi-Gram based decoder for SMS Lingo. In last few years, a significant increment in both the computational power, storage capacity of computers, and the availability of large amount of bilingual data, have made possible for Statistical Machine Translation (SMT) to become a feasible and practical technology. Natural Language Processing is capable of converting almost every machine to a human being by applying artificial intelligence and smart decision making features. In our previous work we employed Bi-Gram Language Model (LM) with a SMT decoder through which a sentence written with short forms in an SMS is translated into long form sentence. This helps users to combine multiple languages with larger vocabulary and is a useful tool for small devices like mobile phones. Since then technology have moved forward with rapid pace and NLP has become the inseparable tool for human machine interface. The IOT platform is taking the world to a place of humans as living machines and appliances as just non living smart machines. In this paper we are discussing some state-of-the-art N-Gram based decoding techniques for text to emotion extraction. The proposed work in the paper is based on outcomes, methods and generated results from various algorithms suggested in different research papers. We have taken the research work one step further and integrated the results with electronic systems on IOT platform which can be controlled and manipulated with human emotions. There are four basic methods to detect emotions from text: Keyword based detection, learning-based detection, lexical affinity method, hybrid detection. From the extracted messages we plan to develop an emotion corpus and then use the time stamped information of mood swing to control devices and digital environment as per expected emotions or mood. So, we are suggesting N-Gram based Smart Living Machines on IOT platform. The emotional intensity of an individual for given circumstances varies from person to person and even time to time for the same individual, hence personalized time stamped corpus for every individual is anticipated. Proposed idea is an IOT controlled system where devices are controlled according to individual’s mood.