Skip to main content
Humanities LibreTexts

3.1.2: Disambiguation by Machine

  • Page ID
    21966
  • \( \newcommand{\vecs}[1]{\overset { \scriptstyle \rightharpoonup} {\mathbf{#1}} } \) \( \newcommand{\vecd}[1]{\overset{-\!-\!\rightharpoonup}{\vphantom{a}\smash {#1}}} \)\(\newcommand{\id}{\mathrm{id}}\) \( \newcommand{\Span}{\mathrm{span}}\) \( \newcommand{\kernel}{\mathrm{null}\,}\) \( \newcommand{\range}{\mathrm{range}\,}\) \( \newcommand{\RealPart}{\mathrm{Re}}\) \( \newcommand{\ImaginaryPart}{\mathrm{Im}}\) \( \newcommand{\Argument}{\mathrm{Arg}}\) \( \newcommand{\norm}[1]{\| #1 \|}\) \( \newcommand{\inner}[2]{\langle #1, #2 \rangle}\) \( \newcommand{\Span}{\mathrm{span}}\) \(\newcommand{\id}{\mathrm{id}}\) \( \newcommand{\Span}{\mathrm{span}}\) \( \newcommand{\kernel}{\mathrm{null}\,}\) \( \newcommand{\range}{\mathrm{range}\,}\) \( \newcommand{\RealPart}{\mathrm{Re}}\) \( \newcommand{\ImaginaryPart}{\mathrm{Im}}\) \( \newcommand{\Argument}{\mathrm{Arg}}\) \( \newcommand{\norm}[1]{\| #1 \|}\) \( \newcommand{\inner}[2]{\langle #1, #2 \rangle}\) \( \newcommand{\Span}{\mathrm{span}}\)\(\newcommand{\AA}{\unicode[.8,0]{x212B}}\)

    What sort of instructions do you give a computer when you want it to disambiguate by being sensitive to context? You need to teach it to use the appropriate background knowledge encoded in its database. For example, suppose you want to build a computer, let’s call it Watson, to understand English and then to translate English into another language. How would Watson handle these two sentences?

    Time flies like an arrow in the sky.
    Fruit flies like a banana.

    When you read these two sentences, you figured out unconsciously that flies is a verb in the first sentence but not in the second. Much of our understanding of English requires a great amount of unconscious disambiguation of this sort. It would be extremely difficult to program Watson with everything it needs to "know" to do this kind of processing for all possible English sentences.

    To explore this problem further, try to make sense of the following statement:

    The chickens are ready to eat.

    Are the chickens ready to do something, or are they about to be eaten? No ambiguity problem occurs with this grammatically similar statement:

    The steaks are ready to eat.

    When you read this statement, you unconsciously searched your background knowledge for whether steak is the kind of thing that eats other things, and you were then able to rule out that interpretation of the statement. It is difficult to program a computer to do this. If there is ever going to be an artificially intelligent computer program that uses background knowledge to disambiguate, then someone is going to have to instruct it to do all the information processing that is done unconsciously by us humans, who are naturally intelligent.

    In the 1950s, when the field of computer science was beginning, many computer designers and programmers made radically optimistic claims about how they were on the verge of automating language understanding and language translation. The U.S. government was convinced, and it invested a great amount of money in attempts to automate language translation. For example, it funded a project to develop a computer program that could readily translate from English to Russian and also from Russian to English. After years of heavy investment, one of the researchers tested the main product of all these efforts by feeding in the following English sentence:

    The spirit is willing, but the flesh is weak.

    The researcher then took the Russian output and fed it into the machine to be translated back into English, expecting to get something close to the original sentence. Here was the result:

    The vodka is strong, but the meat is rotten.

    As a consequence, the government drastically reduced funds for machine translation.

    These examples of the failure of machine translation show us that ambiguity is a serious obstacle to any mechanical treatment of language understanding.


    This page titled 3.1.2: Disambiguation by Machine is shared under a CC BY-NC-SA license and was authored, remixed, and/or curated by Bradley H. Dowden.

    • Was this article helpful?