The current boom of Artificial Intelligence (AI) and Natural Language Processing (NLP) relies on powerful predictive methods that are efficiently able to discover complex functions that relate observed inputs with known outputs. As effective as they are, we know that such methods are essentially shallow and struggle to exploit or expose any deeper meaningful links between the observed data (the signals) with their origins, meanings and causes. The path to bridging the gap between signals and meaning is very long but some promising and practical steps are being taken by AI and NLP researchers.\nIn NLP, general resources like Wikipedia, DBPedia, WordNet, ConceptNet and domain specific ontologies, such as UMLS (Unified Medical Language System) in medicine, are being used to enrich the semantic layer of NLP in important applications, such as health records summarization, Information Retrieval (IR) from web archives, recommender systems, story tracking in journalism and social media. Narrative Extraction and Story Understanding build on the progress of NLP to automatically obtain a narrational global view from free text.