Readme
Plan: Use http://web.airdroid.com/ to mirror screen Take screenshot and crop https://www.pyimagesearch.com/2018/01/01/taking-screenshots-with-opencv-and-python/ Run through google vision https://medium.com/@tobymellor/hq-trivia-using-bots-to-win-money-from-online-game-shows-ce2a1b11828b Search google for results --test out watson to understand intents? Get entities from question: get wikipedia pages: for each entity for each answer
convert to lower case
lookup \
TABLE:
ENTITY1 ENTITY2
ANSWER1 ANSWER2 ANSWER3
TO DO: For the google results: analyse which question words are in results find the answer which re22turns greatest proportion of question entities in each answer
TO DO: get q entities google result count: which of these celebs named his pet after oprah winfrey? tom cruise oprah -> tom cruise oprah pet 50 cent oprah -> 50 cent oprah pet Fast Track Skills 1/2
TO DO: get entities in question get new entity from wiki Search in the wikipedia for the new entity for each answer
on a standard u.k. plug, what colour is the live wire?
red live wire -> red live wire standard uk plug
answer + object +> andswer + object + other entity
work out PERCENTAGE CHANGE
this doesn't work for Oprah wibfrey
need to score the verb after the answer entity type (if there is a large imbalance between classes). If it does not occur in search results, the search results are not valid
TO DO:
if question asks for numbers [amount, least, most]:
e.g. which of these literary awards offers the least prize money?
convert to google search:
answer "amount of" last-entity
booker prize amount of prize money
extract values
TO DO: PAIR QUESTIONS: IF 1+ ANSWERS CONTAIN '/' and each answer has 2+ words search google for answer pair + question entity (e,g -"jermaine randy jackson 5") look up answers in google pages. for each result: calculate the percentage of that pair against the other answer entities for each result: e.g. an article on jermaine and randy without mentioning any other jackson's = 100% Which answer pair has the distribution of results most skewed towards that pair (without the others)
TO DO: get entities in question swap with wikipedia to get new question entity: Best picture Oscar --> Academy award for best picture https://en.wikipedia.org/wiki/Academy_Award_for_Best_Picture Get tables from wikipedia for the wikis associated with each possible answers ---> Filter all tables by ones which have the new question entity in header< find rows which have the new entity
Intersting questions: which pair swapped places when the jackson 5 became the jacksons?