A mini project on Named Entity Recognition, that takes 4 inputs: first and last name, the location and organisation and finds the various mentions of the entity in the web through a google search.

• requests
• sys
• bs4
• nltk
• re
• word_tokenize
• pos_tag
• conlltags2tree
• tree2conlltags
• pprint

• Before executing, uncomment the lines that download punkt, averaged_perceptron_tagger, maxent_ne_chunker, words.
• To run, type python NER.py

Each and every sentence is evaluated by tokenization and adding position tags to every word. The obtained array is converted into noun-phrase chunks with a regular expression. Our chunk pattern consists of one rule, that a noun phrase, NP, should be formed whenever the chunker finds an optional determiner, DT, followed by any number of adjectives, JJ, and then a noun, NN. IOB tags are the standard way to represent chunk structures. After converting, we use nltk's ne_chunk() function to perform classification and addition of categories PERSON, ORGANISATION, LOCATION etc.