Improve lexicon and ontology_manager.py

Currently ontology_manager.py using a pattern match to find words potentially in a lexicon/ontology. You might want to have an exact dictionary to do a lexcion match, like "Barack Obama": "PUBLIC_FIGURE". Also, we need to incorporate better stop words and language specific stopwords to prevent the infering a word belongs in the lexicon when it's a stop words (like "maybe", "must", etc.)

Start by reading ontology_manager.py and trace the current pattern-based lexicon or ontology matching. Determine where exact dictionary matching and stop-word filtering belong; done means matching supports entries such as "Barack Obama": "PUBLIC_FIGURE" and excludes general and language-specific stop words such as "maybe" and "must".

Written by the indexing model from the issue text.