Deriving Semantic Meaning From Google Results

prostoalex writes "New Scientist talks about Paul Vitanyi and Rudi Cilibrasi of the National Institute for Mathematics and Computer Science in Amsterdam and their work to extract meaning of words from Google's index. The pair demonstrates an unsupervised clustering algorithm, which 'distinguish between colours, numbers, different religions and Dutch painters based on the number of hits they return', according to New Scientist."

Comments

The elephant in the living room.

[Eunuch]

These kinds of articles never seem to get a very basic problem--natural languages. English is full of words that trip even humans. "Right" the direction versus "right" the judgement is a good example. In wartime something as simple as that may have lead to death. It's the elephant in the living room. Huge, important problem that nobody wants to talk about. There are alternatives, such as lojban which can be parsed like any computer program.

The article mentions English-Spanish translation. When one language is ambiguous (from a bit of Spanish I had in HS I'm guessing English is far more ambiguous), there is no hope of easy translation. And it's worse because the bigger application may be translating the many English pages (ambiguous) to Spanish.

Re:The elephant in the living room.

[freralqqvba]

Well obviously the technology is not perfect yet. However, none of the problems you bring up are particularly insurmountable (as long as you aren't excepting the AI to be BETTER at parsing languages than people). Yes, words are ambiguous, and yes humans can fail at parsing them, ergo computers probably will too. That's just a fact, we're not going to achieve perfection. Still, this could be a pretty major step forward (well, not that this is the first time something like this has been tried - but the base premise seems sound) by using google the elephant of a problem you mention can be partialy mitigated. Google gives enough context around a word that ideally, when the word to be translated is also surrounded by context its meaning amoung alternate meanings can be discovered without giving an overly ambigous translation.

Re:The elephant in the living room.

[ericbg05]

The article mentions English-Spanish translation. When one language is ambiguous (from a bit of Spanish I had in HS I'm guessing English is far more ambiguous), there is no hope of easy translation.

Every language has "ambiguity", but ambiguity can come in different flavors (phonological, morphological, syntactic, semantic, pragmatic). Some of the chief instigators of language change can be thought of as ambiguity on these levels. So firstly, it's hard to imagine the existence of a function mapping languages to "ambiguity levels".

The motivation for your comment about English versus Spanish probably comes from the fact that you know of more English homophones than Spanish ones. Indeed, most literate people think of their language in terms of written words, so your take on the matter is common.

(As a slight digression, your example of right the direction versus right as in 'correct, just' is pretty interesting. We can understand the semantic similarity between the two when we notice that most humans are right-handed. Thus it is extraordinarily common, cross-linguistically and cross-culturally, for the word meaning the direction 'right' to have similar meanings as dextrous, just, well-guided and so on, whereas the word meaning the direction 'left' also has meanings such as worthless, stupid. (In fact, the word dextrous was borrowed through French from the Latin word dexter meaning 'right, dexterous' or dextra meaning 'right hand'.) So the given example is one where, historically, a word had no ambiguity, but gained ambiguity because speakers started using it differently.)

Getting back to the main topic, more problematic about Section 7 of TFA is the implicit assertion that, at some point in the future, their techniques can be applied to create a function mapping words in a particular language to words in another language. Anybody who has studied more than one language has seen cases where this is difficult to do on the word-level. For instance, the French equivalent of English river is often given as riviere or fleuve. But riviere is only used by French speakers to mean 'river or stream that runs into another river or stream' whereas fleuve means 'river or stream that runs into the sea'. English breaks up river-like things by size: rivers are bigger than streams. So, in the strictest sense, there is no English word for fleuve, just as there's no French word for stream (unless there has been a recent borrowing I don't know about). This certainly does not imply that French people can't tell the difference between big rivers and small rivers; their lexicon just breaks things up differently.

These little problems can be remedied lexically, as I've just done. So fleuve is denotationally equivalent to river or stream that runs into the sea, although the latter is obviously much bulkier than its French equivalent. The real problem is that there are words in some languages whose meanings are not encoded at all in other languages. English, for example, has a lexical past-progressive tense marker, was, used in the first person singular (e.g. I was running to the store). Some languages have no notion of tense. What, then, does was mean in the context of such a language?

It's pretty well-known that Slashdotters' general policy is to tear apart every article we read, and half of those we don't. This is certainly not my intent here. Languages are complicated beasties, and everyone seems to understand that, including the writers of the article. So, we should interpret their result in Section 7 as them saying, "Well, maybe this has gotten us a baby-step closer to creating the hypothetical Perfect Natural Language Translator, but someone's gonna have to do a lot more work to see where this thing goes".

Re:The elephant in the living room.

[danila]

Well, I don't think there is a way to translate a text into another language without

1. understanding the text
   
2. understanding both languages
   
3. understanding the socio-cultural context of both languages

But we must consider the fact that most humans can't produce a decent translation either, even if they think they understand both languages. I've been professionally translating movies (EN->RU) and I know to what extent the scripts are riddled with linguistic traps. An average professional human translator would be lucky to produce a 90% valid translation (much less a perfect one).

So if the developers of computer translation tools do not strive for perfection, but for an average human level, they might succeed rather soon, by using a combination of several approaches (including the one described in this paper). Of course, a translation program can confuse NATO with the Northern Alliance, or Thai with Tahitian but so can a human.

Semantic meaning?

[zorren]

I though semantic meant "meaning".

Re:Semantic meaning?

[exp(pi*sqrt(163))]

Meaning could, in principle, mean 'affective meaning' as in the emotional weight something carries. Maybe Google are also working on emotional search engines and the article poster doesn't want us getting confused with that.

Scientology

[Jace of Fuse!]

Is this in any way related to the way that Google was able to decide all on it's own that Scientology was crap, and thus bring Operation Clambake up to the top of the search results? (Until they Scientology people got pissed, anyway.)

Google is already starting to show signs of intelligence higher than some people. :)

Extend this to the library of congress...

[physicsphairy]

While I think ideally you would endow computers with the same algorithmic usage of speech that is employed by human beings, as these researchers have shown, it is also possible to work with programs that do not 'parse' language but rather categorize it based on massive databases of language that has already been parsed by humans.

This obviously has its failings, but theoretically, you could use a sufficiently large database of common human language coupled with simple algorithms to perform operations like grammar checking.

An internet search would not be quite so useful for that, but I would really be interested in what would be possible with full digital access to the library of congress. I would imagine you could do things like automatically generate books based on existing material.

Would that be 'semantic meaning'...

[exp(pi*sqrt(163))]

...as opposed to 'non-semantic meaning' or just 'semantic meaning' as in 'I don't know what semantic means but using it here will make me look intelligent'?

Compression is a stricter test for AI than Turing

[Baldrson]

From the linked academic abstract:

Viewing this mapping as a data compressor, we connect to earlier work on Normalized Compression Distance.

This is basically what I was referring to in my response to "Using The Web For Linguistic Research" when I said:

There needs to be an annual prize for the highest compression ratio using random pages from the web as the corpus. This would probably do more for real advancement of artificial intelligence than the Turing competitions.

followed by the explanation:

Intelligence can be seen as the ability to take a sample of some space and generalize it to predict things about the space from which the sample was drawn. The smaller the sample and the more accurate the prediction, the greater the intelligence. This is also a short description of what a compression algorithm does.

and

Text Compression as a Test for Artificial Intelligence, 1999 AAAI Proceedings. Matt Mahoney shows that text prediction or compression is a stricter test for AI than the Turing test. (1 page poster, compressed Postscript).

Good for scholars, bad for geeks

[kyndig]

This is a pretty nice approach. Quoted from the news article "The technique has managed to distinguish between colours, numbers, different religions and Dutch painters based on the number of hits they return, the researchers report in an online preprint.", it shows that common terminology can be drawn. In the end though, this is a refined search routine for Google IMHO. This would be good for scholar searches perhaps, or even a dynamic thesaurus. But when using terms such as: does windows use linux, the derived results would be broken down into: "linux" "windows" "use" . Google cached pages containing these terms vary so greatly in content. But, if searching for something along the lines of "dutch painters favorite colors", would produce desired results like the control method used in the news article

Re:wARTIME?

[MoonFog]

Well, when I was in the army, it was very strict that whatever was said over a network DIDN'T have an ambigous meaning. That's why the army language sounds kinda weird at times, because you are not supposed to misunderstand anything.

not many will get this

[2TecTom]

First off, I am not an "AI" expert nor do I claim to be, however, this is how I see it.

Since it seems that so few really understand the term "intelligence", it is really not surprising that even fewer grasp the meaning of the term "artificial intelligence", is it?

One: intelligence is not awareness.

Although we cannot prove the existence of or even seem to really define self-awareness, it seems self-evident, at least to me, that intelligence is clearly defined and can be measured.

Therefore, I believe that we will have "artificial intelligence" soon, in fact, I'd bet Google may well be the first AI or "self intelligent' engine.

However, I suspect it will be quite awhile before we are mature enough to build a self-aware engine.

Lastly, in regards to some of the other comments, it seems to me that this paper is about using the "intelligence" included in the language we use, that Google crawls. This repository is the single largest collection of semantic weighting, therefore, algorithms could be developed that reflect this "intelligence", therefore appear themselves intelligent, even though they themselves are simply deterministic.

Whew ...

Unsupervised but Reflective of Human Preferences

[reporter]

Even though I disagree with Google's hiring practices (i.e. preferring H-1Bs when many American engineers are unemployed), I must admit that Google's search algorithm is the best one -- even better than Yahoo! Search, which I use regularly for socio-political reasons.

I will give you an example. If you search news (i.e., either Google News or Yahoo! News) for stories about the recent federal action (by Washington) involving Chinese companies and Iranians weapons improved by Chinese technology, you will discover that one of the popular news articles about this topic comes from the "New York Times". Several other newspapers redistributed the Times article, written by David Sanger (spelling?).

I read that article, but I also read articles from less popular Web news sites: e.g. "Taipei Times". The "Taipei Times" article does mention that a Taiwanese company was also implicated in the sale of weapons technology to Taiwan. Yet, "New York Times" article made no mention of this fact.

Is the "Taipei Times" telling the truth? It claims that Ecoma Enterprise Company, a Taiwanese company, was one of the culprits.

At this point, I fired up both Yahoo! Search and Google. Only on Google was I successful in locating the the ORIGINAL source of the information about American penalties against the 7 Chinese companies and the 1 Taiwanese company. The information is on page 133 of the "Federal Register" (volume 70, number 1). So, I discovered that the "Taipei Times" was telling the truth.

Guess how long I took on Google to find this information? 5 minutes. I kid you not. Even though I hate Google's employment practices, I am quite impressed with their technology.

Using Yahoo! Search, I was not able to locate the desired information.

Apparently, Google has an algorithm that, although it is unsupervised (i.e. without the kind of human interaction that corrupts Yahoo! Search), it captures the notion of what the typical person wants to find. The Google algorithm, dare I say "it", is on the verge of acquiring human sentience. THAT is, indeed, impressive.

Pray to Buddha that the middle name of the CEO is not "666" or Beelzebub. Just kidding.

Limitations of NGD (Normalized Google Distance)

[G4from128k]

Although very clever, NGD (Normalized Google Distance) misses alll higher-order relationships and does not even distinguish between different categories of pairwise relationships. For example, NGD might assume that "Bush" & "Iraq" had the same relationship as "Slashdot" & "Geek" because the two word pairs co-occur with similar frequencies.

More interesting are analyses on n-Tuples (co-occurences and orderings of n-words at a time). Anyone who does ER (Entity-Relationship) diagrams for relational databases will appreciate that many relationships involve multiple entities that are decomposable into pairwise relationships.

Another limit is that Google is atrocious on its estimates of the number of hits. The actual number of hits is only fraction (about 60%?) of the estimated from my experience. This suggests that Google has a pairwise estimator built in that may be only partially empirical. If Google simply reports an estimated number of hits based on products of probabilities, then their is no information about the pair in the NGD. Obviously, these scientists have gotten useful results, but NGD may not be as good an estimate of the co-occurence of the words as the scientists assume.

Re:Limitations of NGD (Normalized Google Distance)

[Rudi Cilibrasi]

You are right that Google may be performing estimation and this could effect results and I don't really know what sort of rounding they do at this time. Perhaps more will become apparent. But your other assertion about no higher order statistics is incorrect. see the earlier Clustering by Compression paper for more info. Quickly, the reason is as follows:

• I use NGD to convert arbitrarily-large lists of search-terms into feature-vectors of arbitrary dimension. The only limit to this is the max query length for Google, and this is just a detail.
• I use a Support Vector Machine with a Radial Basis Function kernel. The RBF kernel has an effectively infinite dimension and so can learn any function. SVM is a universal learner like neural nets and many other famous algorithms. So higher-order features (composed of products of several NGD) can indeed be used in learning.

The main purpose of the research is in extending generality of automatic learning. See the earlier papers in the series including Algorithmic Clustering of Music, and the earlier theoretical work. NGD is a special case of NCD. NCD is a family of functions that can be used as the basis of a universal learning system in a variety of ways. Our theory justifies this innovation and leads to a whole class of easy to write algorithms.

Thanks for your interest, it is good to see that this research is striking a chord with the Slashdot community. I hope this leads to a whole lot of more easy-to-use semi-intelligent software. Cheers!

Re:Huh?

[modifried]

Google has determined the meaning of life: http://www.google.com/search?hl=en&q=42&btnG=Googl e+Search

I think you might've meant:
http://www.google.com/search?hl=en&q=what+is+the+a nswer+to+life%2C+the+universe%2C+and+everything%3F

Been working on similar

[Arngautr]

I wrote a program that gathered, analyzed and used word pair frequency data (various situational pairings). It needs more raw data, but shows a lot of promise. I opted to not use literature, as that often has archaic and purposefully awful word usage. Some of the issues involved include case, like Fall vs fall, I chose to ignore case, grammatical structure, needs to integrate with a grammar checker. Coupling this with a thesaurus is my eventual goal, this leads to some obvious difficulties, though it has potential rewards. I had considered google, and have run a few tests using it, but that solution was too simple, and not quite as powerful in the long run. Just had to share, sorry to waste your time.

Re:wARTIME?

[Fjornir]

Er. You didn't get the joke, so I will explain. The lore is that "repeat" is a command to the artillery to fire again on their last target, so you never ever say "repeat" on the radio, instead you say "say again".

The lore also contains an interesting anectode about the '92 riots in LA. Apparently a group of Marines were dispatched to assist the police. Two officers were approaching a house when someone opened up with a shotgun at them. One officer shouted "cover me" -- so the Marines proceeded to lay down covering fire on the house -- more than two hundred rounds were fired into that house.

Language is more than words

[Hal XP]

English is full of words that trip even humans. "Right" the direction versus "right" the judgement is a good example.

"Right" isn't really a good example of a word that might "trip even humans." A human (translator) will parse not just by word but will attempt to extract a word's meaning from the surrounding phrases, sentences or even paragraphs. The syntax of the language may also come into play. In spoken language, additional "clues" can be derived from the situation in which the word is spoken, and often the extra-textual "body language" is more important, e.g. a hand pointing right or a head nodding in approval. I don't think an adult would be confused by the sentence "You're right. Let's go right." In wartime, I can imagine a responsible English-speaking commander barking references to GPS locations or using body language. It would be a mistake to think of a word in isolation from its context. After all, even in computer languages, a printf or goto by itself will chuck off a compiler error.

Re:Language is more than words

[Hal XP]

There ought to be a military regulation forbidding the use of anything other than "Yes," "No," or "I don't know, sir" in a combat situation. Right?

Limits to semantic derivations from Google

[saddino]

My company develops a data mining program for OS X (theConcept) that uses Google (or other search engines) to provide links to data for mining.

For example, searching on Google for "tom cruise" brings up pages upon pages of links, but -- from a cursory glance at the results -- it is impossible to learn anything about Tom Cruise unless one visits those results.

Our software visits each of those results (for example, the first 100) and looks for the most significant keywords and phrases used over all the data. As you might expect, these typically end up being the names of people (e.g. Nicole Kidman, Penelope Cruz) or movies (e.g. Top Gun, Color of Money) that are associated with Tom Cruise. As far as our software goes, this is ample for doing keyphrase analysis.

But the problem with deriving any additional meaning from the Internet web space is this: the biases that exist due to the very reasons for mentioning Tom Cruise (namely those things he is famous for) simply outweigh -- by a wide margin -- any other quite relevant interesting data about Tom Cruise. So, in fact, the web, in general, is an awful corpus of valid semantic data.

If you want a rough model of popular ideas then perhaps Google and the web en masse is useful (it is for our software). But if you want any real meaning at all you come to the same conclusion that has given rise to sites like Wiki: the web, to be blunt, has a whole lot of shit in it. Coming up with a perfect (and rational) filter is quite a task.

On the bright side...

[Anonymous Coward]

They are developing an open source tool http://complearn.sourceforge.net/ that will hopefully integrate the algorithm they describe. Right now it's only supporting one of their previous algorithms. More about this in the above link and section 5 of the paper.

Pretentiously titled

[Turadg]

I've perused the abstract and skimmed the body of the paper. They're fine. But the title is misleading: Automatic Meaning Discovery Using Google.

Their software has discovered meaning no more than paper has when the lexicographer is done writing her dictionary. Meaning is not the grouping of symbols.

For systems that step towards encoding meaning as human brains do, consider the Neural Theory of Language.

What you build is a substrate.

[Dylan Thomas]

You're quite correct that cowboy-loose definitions of terms make this a very difficult discussion to have. For example, when you say "self awareness," it's unlikely that you actually mean "self awareness" in the literal sense; after all, if a computer is capable of detecting when its processor is overheating (and perhaps turn on a fan in response), it is basically "self aware," though we wouldn't confuse that with itelligence.

Rather, I think by "self awareness" here you mean, possessing narrativity; that is, the ability to construct a narrative of itself in relation to the things of which it is aware. In simpler words, consciousness. Now, it is possible to be intelligent without being conscious (everyone thinks they have the smartest dog in the world, but that doesn't make the poor beast conscious). But is it possible to be conscious without being intelligent?

Consciousness is fundamentally linguistic in origin (and I'm tired of arguing that point with people who haven't done a day of cognitive studies in their lives; there's no way around it: without language, consciousness does not evolve). So, for example, in the course of human evolution, first a linguistic parsing system was evolved, humans got language, and then, once this substrate was established, consciousness evolved as an epiphenomenon which rode on top of it. This substrate proved to be a fertile breeding ground on which memetic evolution could take place, as well, and since that is broader than any one particular human component in the system, it's almost more proper to say that we are the tools memes use to propogate, and not vice versa. (This argument is fairly well established with genes; same rules apply.)

So, any artificial system which contains "consciousness" will have to first handle language. If you don't have that linguistic substrate for narrativity and memetic evolution, there is nothing for consciousness to occur in. Maybe the information is there, but it would be like me pointing to an empty spot in the room and saying, "That's a balloon full of air; I just forgot the balloon." So, let's do this in the proper order: language first, then consciousness.

understanding relationships is intellegence

[menem]

If given perfect information about the relationships between concepts, you could derive a very intellegent machine. TAke a human for example..

A baby hears the word mom spoken by his mom. Gradually, the baby knows there is a relationship between that sound and a smily face.

The child, growing up, starts to see relationships. Intense pain, which is rare, when correlated with a hot stove, has strong meaning in his mind.

Everything is learned initially through correlations. The advantage of human beings is that there are many more data points for correlation. Google's correlations are weak and don't give nearly as much information.

No.

[Dylan Thomas]

A slug is not conscious. Nothing without langauge is. Recommended reading: Dr. Daniel C. Dennett, Consciousness Explained and Darwin's Dangerous Idea. Richard Dawkins, The Extended Phenotype. Julian Jaynes, The Origin of Consciousness in the Breakdown of the Bicameral Mind.

Those are all more commercial works, well within the grasp of even people who've done no work in the field. For more sholarly and technical references, check their bibliographies, especially in Dennett.

Down, cowboy.

[Dylan Thomas]

"Sheesh" is a word which normally means, "I'm not very good at actually saying what I mean, so I'll just make strange noises and roll my eyes at someone who won't figure it out for me." (It's also the nick of one of my favorite Internet trolls; what ever happened to the good old days when trolls actually tried to be entertaining, instead of merely annoying?)

Anyway, okay, it's loose definitions of words that are once again getting us into trouble here. That a slug is aware of its environment, as in, capable of responding to environmental stimuli, okay, I'll give you that one. I won't gift wrap it for you, but I'll give it to you.

But that's entirely different from "consciousness" in the sense that we're discussing here. After all, even a computer is capable of detecting environmental stimuli, and responding to them, but as my colleague 2TecTom is pointing out in this same thread, the mere ability to respond to environmental stimuli is not synonymous with consciousness.

Read the source material. It'll give you the weapons you need to overcome your sheeshing.