Looking for a paper comparing NLP methods with Data Mining techniques

I've recently attended a conference, where one of the participants mentioned a recent paper published by a Google employee, which showed that using data mining techniques in application to NLP might sometimes give better results than using the NLP methods themselves.

That was the only description he gave and on top of that, he didn't remember author's name.

Is anyone familiar with such paper?

there are probably several papers which fit this criteria and this may not be enough info/ criteria to really narrow it down. however some general bkg: google has pioneered research into the use of statistical algorithms for NLP for years and particularly (somewhat more recently) machine auto language translation. in this research one has a large database (aka "big data") of "translated sentence pairs" between language A and language B and the machine learning algorithm attempts to learn the relationship between the two, and its considered something of a relatively recent paradigm shift in the field because previously it was not considered/ thought possible, but google has achieved very good and improving results in the area, and my understanding is that some of this is now used/ incorporated in their live/ production/ public "translate" feature/ "product" built into the main search engine results.

try this nice list of papers: Natural Language Processing listing most NLP papers/research at google by date. there are 8 papers with the word "statistical" in the title. the most recent paper from 2013 may be close to what is being identified or cite the nearby area your speaker was referring to. other papers are from 2012 or older.

• One Billion Word Benchmark for Measuring Progress in Statistical Language Modeling / Ciprian Chelba, Tomas Mikolov, Mike Schuster, Qi Ge, Thorsten Brants, Phillipp Koehn, Tony Robinson

ArXiv, Google (2013)

StackExchange Computer Science Q#35900, answer score: 3