Teaching Learning Algorithm

One thing is for sure; computers learn a heck of a lot faster then humans do. One bumper sticker I love is "I would rather be writing programs that write programs." I'm something of a specialist in this fundamental task. Reason being is I'm totally lazy. For instance, once I had to populate an array of all cities and states into a static array in Java. Instead of manually writing code line by line, I simply wrote a program that wrote the lines of code for me. A little copy and past and compile actions and I was on my way in 8 minutes. So to teach my learning algorithm I simply do as I did before. I automate the process of data entry with a computer to teach itself. To do this, I am entering the strange new world of Web 3.0 where everything is magically connected together. To do this I am simply going to Mark Zuckerburg it and scrap data from provided search engines. Objection! You cry out "You must populate the fields of the search engines with every entity your searching for." Like apple, there is an app for that too. We simply write another program that writes a programs that populates our searches with data from our database. A little vector space analysis and Solr and we got yourself a steaming hot Corpus that has been populated from current events in the world. Apply our learned results to our learning algorithm and your algorithm now has an advanced degree in current events and natural language processing. You may ask "Well why dont I just use Wordnet or New Oxford Dictionary?" Been there, done that, sucks at Word Sense Disambiguation (WSA). Eat that buzz word. Example: "Going downtown Paris tonight." You in I could distinguish Paris as the city of love. A machine may distinguish it as the party chick. Someone with their mind in the gutter may assume the same about the sentence. Since we did VSA on our results earlier, we have the top 100 words with the highest weight. Take the context of the sentence, hopefully an article we Googled earlier mentions "Downtown" and "Going" so we can map those terms to our lexemes in our database.