In the beginning, it was all about the algorithm. PageRank and its “no humans involved” mantra dominated Google since its inception. In recent years however, Google has started to expand the role of “conceptual” knowledge in different areas of its services. The main search bar and its capacity to do all kinds of little tricks is a good example, but I was really quite astounded how seamless concept integration has become on my last trip to Google Translate:

Post filed under algorithms, ontologies, search engines.


  1. Interestingly, I get “words and things” as the first suggested translation. Perhaps because I have English-language settings? Anyway, I wanted to point out that this is not necessarily evidence of an underlying ontology or conceptual knowledge. Google’s translation algorithms rely on statistical correlations found in parallel texts, and it’s not surprising that a correlation would exist between the phrases “les mots et les choses” and “the order of things,” perhaps an even stronger correlation than between the French phrase and its literal English translation.

    • Bernhard says:

      In fact, “les mots et les choses” gets “words and things” while “Les mots et les choses” gets “The Order of Things”. And yeah, it’s true, it’s probably just statistics – I thought so too after posting and I tried out a couple of other titles: “Les spectres de Marx” gets “The specters of Marx” while “Les spectres de Lénine” gets “The spectra of Lenin”. For that kind of statistical precision they certainly work a lot with Google Books and different language versions of the same book. Better than an ontology!

  2. Hi.
    Maybe relation between the phrase and its translation is made just by statistical methods, but the conceptual relationship between sentences can be reached by statistical methods mainly if the occurrences are obtained from the users behaviour. You can generate an ontology (not perfect because the users can make mistakes).

Leave a Reply

Your email address will not be published. Required fields are marked *

Tech support questions will not be answered. Please refer to the FAQ of the tool.