In case if you are looking for a reliable solution for entity extraction, then the RL3 engine should be in your shortlist for sure (BTW, is free for non-commercial use!). It is actually based on regex but allows named patterns, modules, templates, and kind of classes. Should you be OK doing some coding and understand NER & NLP basics, lots of helpful staff await you under the hood. For example, some ready-to-use pattern libraries, possibility to reuse patterns by adding them into a custom library, compatibility with third-party dictionaries of enormous size (in the form of text files).
Also, I have had an opportunity to play with an RL3-based web scrapping service which crawls corporate websites (~25 million) and extracts key information such as people, contacts, etc. It is fast, and RL3 patterns work very precisely.
