How to detect the events using NLP and Machine learning? -

i have text describing events such birth , new job , wedding , death etc .. or no event . how detect these events ?

my approach form set of words , search them in text corresponding event . or use bayesian classifier . bayesian classifier requires training on classes , require method can classify without giving types of examples ? possible ?

what performance requirements? low recall ok? need high precision?

based on question, i'm guessing want reasonable recall (reading classified ads send out spam?) don't have training data.

you want method can classify documents (using events named entity recognition algorithm features) without providing training data. supervised methods (including bayesian ones) require training data, asking/wishing not possible. need labelled data in case, otherwise how can tell how detection process working?

at stage should not worried classifier use - suggest writing handful of regular expressions see how hard problem / performance get. may dozen regular expressions may 90% of these events , can avoid over-engineering problem. luck!