There are two quotes that I remember from my CS profs that have stuck with me over the years.
AI prof addressing statistical vs. rule-based learning: "All models are wrong. Some models are useful."
Systems level programming prof: "Now that you have a thorough understanding of programming all up and down the stack, if you want to minimize the amount of money you make and maximize the hours you work, then take a job as a video game programmer."
I know several coders who have worked in the gaming industry for 5+ years and they're all completely jaded at this point.
I remember this happening 5 years ago, but instead of a bunch of salesmen frantically searching for businesses to pitch couponing to, it was mortgage brokers pitching exotic loans to home buyers.
I'll go out on a limb and guess that most of Groupon's sales team consists of ex-mortgage brokers and washed-up real estate agents.
Dvorak argues that Facebook is AOL. I disagree, and think it's inferior to AOL. Mainly because Facebook neither provides an entry into the internet (like AOL's ISP did/does) nor has a viable, general purpose email system. Facebook messaging couldn't replace something like Gmail. I think the lack of those systems make it easy to abandon the service for something else.
Having 100+ "friends" on Facebook isn't a compelling reason to stay when most people probably wouldn't notice if 90% of those individuals disappeared from one day to the next. Most of your social network online is an illusion.
I'm testing out Google+ right now, but I'm doing so under an alias. The alias is known to my friends and everything I post and share is authentic. My change in attitude comes from 4 years of having a real identity presence online; I don't really see the benefit of shedding anonymity.
Spammers increase conversions and activity on the site. I worked for a dating site and they condoned spammers that weren't out of control. Hell, they tacitly approved of hookers working the site as long as they weren't blatant.
If you're looking for a date, take the effort of going out and meeting someone. Online dating is a total waste of time for most people, unless you're looking for someone who sits at their computer all day.
That sounds possible, but gathering all the data for topic-specific training might be somewhat maddening. The problem you get with larger groupings of words is lack of cohesion. If you trained it at a sentence level, you might be able to produce coherent sentences. But as you generated more sentences to produce a paragraph, it would likely meander.
I created a kind of Mad Lib generator using CFGs. A paragraph consisted of: [Intro] [Supporting sentence 1] [Supporting sentence 2] [Supporting sentence 3] ... [Conclusion]. All the sentences had placeholders for various nouns and adjectives that could later be filled in programmatically, and I extended the grammar spec to both permute sets of sentences and generate null productions with certain probabilities.
The base sentences were created by humans, about 15 per grammar rule. A single person could create a topic-based paragraph/grammar in less than two work days. The chances of it creating the same template twice was about one in a billion. Of course, the probability varied depending on how many seed sentences were present.
If the person writing the seed sentences is literate and passed the 6th grade, then everything the program generates is indistinguishable from human text.
That's interesting, do you have any code or examples?
My thought about topic detection is to have it learn which words go together, and then augment the Markov chain model by some method that would weigh the Markov chain probability with the topic probability to select the next word, so it would at least generally stick to topic-relevant words. Perhaps you could even select one topic (a sample sentence, really) in the beginning and have it generate sentences based on that for the entire document.
I wish I could publish it, but my company isn't very much into open source. It's a standard context free grammar framework modified to generate output in a stochastic manner. So it's basically a stochastic context free grammar (SCFG). I can go more into depth in private if you like.
The phrase for finding word pairs in text corpora is "cohort analysis". I was a on research team that did studies of that; mostly finding them, not generating anything with them.
Given Rocky's claim that investing a large amount up front for a traditional advert placement is equivalent to receiving a short-term loan from Groupon for running a deal on the site, then Groupon is merely like all other direct marketing/direct response operations. So you can probably predict how successful Groupon could be by comparing it to something like Valpak.
Despite the fact that Groupon shifts the marketing costs from the merchant to the customer, it probably won't affect couponing behavior in the long run. To me this is like chess where the players have switched sides after a match. It's the same game, but a new player gets the first-move advantage this time around.
Groupon Now! seems slightly more interesting and possibly has more potential.
My wild, unsubstantiated prediction is that they'll IPO, fizzle out and be bought out by some media/new media conglomerate by 2014.
AI prof addressing statistical vs. rule-based learning: "All models are wrong. Some models are useful."
Systems level programming prof: "Now that you have a thorough understanding of programming all up and down the stack, if you want to minimize the amount of money you make and maximize the hours you work, then take a job as a video game programmer."
I know several coders who have worked in the gaming industry for 5+ years and they're all completely jaded at this point.