Towards a technical sentiment lexicon for the maintenance of human-centred buildings

Following the “Cognitive Building” concept, in a few years, building automation systems were drastically improved to collect large amount of user data. However, despite this evolution and research efforts field, human-building interaction remained one least mature fields science due occupants’ complexity diversity. Collecting data became simple cheap, but transforming collected “data” into valuable “information” able create an effective between buildings occupants remains complex. This work contributes by proposing method translate unstructured data, coming from Computerized Maintenance Management System (CMMS), information useful improve interactions management maintenance process. End-users’ requests, through CMMS, used technical sentiment lexicon predict priority intervention based on inverted naïve Bayes approach. Sentiment lexicons are part analysis, interesting field introduced study people’s opinions, sentiments, emotions, attitudes Natural Language Processing (NLP). The is immediately perform assessment contemporary end-users’ thus being more rapid than traditional Machine Learning methods.