Affective computing for smart operations: a survey and comparative analysis of the available tools, libraries and web services

Resumen

In this paper, we make a deep search of the available tools in the market, at the current state of the art of Sentiment Analysis. Our aim is to optimize the human response in Datacenter Operations, using a combination of research tools, that allow us to decrease human error in general operations, managing Complex Infrastructures. The use of Sentiment Analysis tools is the first step for extending our capabilities for optimizing the human interface. Using different data collections from a variety of data sources, our research provides a very interesting outcome. In our final testing, we have found that the three main commercial platforms (IBM Watson, Google Cloud and Microsoft Azure) get the same accuracy (89-90%). for the different datasets tested, based on Artificial Neural Network and Deep Learning techniques. The other stand-alone Applications or APIs, like Vader or MeaninCloud, get a similar accuracy level in some of the datasets, using a different approach, semantic Networks, such as Concepnet1, but the model can easily be optimized above 90% of accuracy, just adjusting some parameter of the semantic model. This paper points to future directions for optimizing DataCenter Operations Management and decreasing human error in complex environments.