When we left off last time we used a search engine API to gather a whole bunch of documents with the term "played Spiderman" or "who played Spiderman." Now we have to process these documents to answer the question. Fortunately, the documents are basically just big strings of text.
Since we're doing a "who" query we want to find all the names in all the documents. This leads to a class discussion on how to find names in a large string. They'll come up with rules like:
- two adjacent words each with a capital first letter.
- an honorific (Mr, Ms, etc.) followed by a capitalized word
- A word that matches a name from a "popular name" list followed by a capitalized word.
The class might come up with different or additional rules. Actually writing the code to pull out the names is an nice little assignment that can be done with classes at many different levels. A beginning class might just use string operations while an intermediate one could use regular expressions.
Will the students write code that always finds all the names? Probably not. Will their code incorrectly identify some word sets as names? Most likely. It doesn't really matter. Actually, this is a good thing. You now have a great platform to talk about false positives and true negatives.
Now the moment of truth - what's the most common name? To answer that you have to decide on the meaning of "most common." Is it the name that appeared the most times? How about the name that appeared in the most documents? Does it matter? Whichever way you do it, chances are the correct answer will be near the top of your list.
Next up in part 3 - other types of queries and why I love this sequence of lessons so much.