Comments Off
(Comments to a blog post this morning on Gizmodo.)
Type in the word “dumb” to Wolfram|Alpha and you’ll get all sorts of interesting information. Like that “dumb” was first recorded in use in 1323—686 years ago.
One thing Wolfram|Alpha gives is synonyms for dumb. An example is “dim”.
But what happens if one looks at the synonyms of synonyms of “dumb”?
Each of the 8 direct synonyms of “dumb” has many more synonyms.
And one can build up a network of all these synonyms of synonyms.
Of course, the network is quite approximate, because one is looking at the synonyms for synonyms, just as words—ignoring the sense of each word.
So, “dim” is a synonym of “dumb”. But what are synonyms of “dim“?
Two of them are “black” and “bleak”. Perfectly reasonable synonyms for “dim” on its own. (See the first definition for dim.)
But now two sensible synonyms lead to something dumb. We get from “dumb” to “dim”, and from “dim” to “black”.
There are 8 first-level synonyms for “dumb”. There are 66 second-level synonyms (of which “black” is one). (A tree would show 102 second-level synonyms, but some are not distinct.) At level three, there are 423
synonyms. At level four, 1851. And so on. (For those who are interested, Mathematica‘s WordData quickly tells us that the sequence continues: 6731, 14800, 21054, 24140, 25459, 26003, 26211, 26292, 26313, 26326, 26330, 26330, 26330—it finally hits the “edge” of the set of connected words.)
By the time one’s even at level three, one’s at words as different as “well-off” and “drapery”.
This is a kind of “Six Degrees of Kevin Bacon” for words. It’s somewhat interesting from a linguistic point of view. But it certainly shouldn’t be used to tell one a substitute word to use in an essay. Imagine substituting “drapery” for “dim”. Or “black” for “dumb”.
The pod labeled “Synonym network” is called that because it gives the network of successive sense-independent word synonyms. It’s always a challenge to come up with short titles for pods that convey all the meaning of their contents.
And in this case one might conclude from “Synonym network” that Wolfram|Alpha thinks “black” is a real synonym for “dumb”. It doesn’t.
Arguably, we could change the title “Synonym network”. At some point during testing it was called “Word association network”. But people didn’t understand at all what that was.
Any suggestions?
I am an avid thesaurisizer and I love the idea of the Synonym network. However, the utility of this feature would be greatly enhanced if the network diagram were actually navigable and more detailed. The basic tree structure is there, but the words at each node are too sparsely labeled.
That’s probably could be done with java, read this:
http://community.wolframalpha.com/story.php?title=interactive-2d-and-3d-graphics
Why not just call it Word Associations? It kinda makes sense just by looking at it
Easy – leave it as it is and add some kind of ellaboration or help button (that either zooms in or explains how the diagram works and hence why W|A is not racist), or show some first level synonyms by default.
Actually (having now looked at the page), a zoom function would be really useful… I’m thinking the kind of scroll-to-zoom, click-and-drag-to-move type interface one sees in their “generic online map” website. (Nice how I didn’t mention Google there)
If people got to the end of your post and understood it, they’ll certainly know what “Word association network” means — although it is a bit ambiguous.
Since not *every* word in the synonym network are synonyms for each other as you state, how about something that describes it, such as “Degrading synonyms network” or “Cascading synonyms network”.
Yes, I like “Cascading synonyms network”.
By the way, it’s nice to see you guys responding intelligently and good, overall, to such an ignorant blog post by the geeks at Gizmodo that… that’s enough on that.
The nun at the british boarding school that I went to, thought everyone other than British were dumb! Ofcourse only way to converse with British was to speak in English, otherwise one is as good as dumb!
It is amazing that there are only half of billion English speaking people in this world! That is 1/13th of the world’s population!
How about Synonym Interface?
I honestly don’t know; to me, this seems to be more of a user issue (user ignorance?) rather than any incongruity on the part of Wolfram|Alpha.
Regardless of how inconvenient the answer may be, the truth remains that there is a viable word path that exists from linking “dumb” (lacking intellect) to “dim” (lacking intellect but also defined as “lack of illumination / light”) to “black” (which, if taken in association with “dim’s” secondary definition: lacking illumination / absence of illumination).
Like i said, I have no idea, that’s just my humble opinion
The antonym for the word dumb is Sarvagna!
In Sanskrit Sarva = All and Gnana = Knowledge. Sarvagna is the one with all the knowledge!
Ofcourse, Latin and Sanskrit are considered dead languages!
I may have missed this, but why are certain words spelled out in the Synonym Network while others just appear as dots?
synonym tree
As you branch out you become further removed from the trunk. Could also incorporate etymological information by exploring ‘roots’
similarity web ?
synonym web?
word-relation(-web)?
or maybe mesh instead of web or stick to the old ‘network’?
However I like the synonym network; because that is what it is..
I think that the problem stems from the fact that a single word can have several meanings. For example, one definition of dim relates to intelligence, whereas another relates to brightness (luminosity?) Perhaps there could be a greater separation between the separate definitions of the word: clicking on a word in a list of synonyms would bring up only the relevant definition, not the complete list of definitions and synonyms. An unaware user would only be able to travel within the web of synonyms with the same definition. Users who wished to view the complete definition for a synonym would be warned that the remaining words are no longer synonyms for the initial word.
On a side note, a Venn diagram would perfectly capture the relationships between synonyms. The main word’s definitions would be represented by overlapping circles, with the main word in the overlap between all the definitions. Synonyms would be shown in the overlap between the definitions they share with the main word. Unfortunately, words with many definitions would create ugly and dense Venn diagrams.
You asked for suggestions for an alternative name for synonym network. I think it satisfactory as it is. I would suggest though that the lines be changed to arrows to show the dirction in which the chart is built. Given the 6 degrees of separation theory for human contacts It would be interesting to know how many connections are needed to connect any two words.
On the ‘Personal Information Page it gives counts of ‘Submitted Stories’ etc.
In keeping with your principle of the power of combining simple elements to form a complex whole I suggest that by clicking on the name against the count a user be shown a list of the items making up the count.from which he can look at any one of them etc.
I think perhaps that the issue is not the title, but the layout. Because it is so extremely sparsely labeled, and navigation consists only of ‘zoom out,’ we cannot SEE that ‘dumb’ leads to ‘dim’ leads to ‘black’. It’s of little immediate interest to me to know that Alpha knows 8 synonyms for ‘dumb’, and that one of those 8 words has a synonym ‘behind;’ I want to know which one. I want to follow the trail. 6 degrees of Kevin Bacon is not much fun, if you don’t get to look in amazement at the chain that leads from one to the other.
I agree with naked ape that stronger navigation would alleviate confusion by allowing the reader to *intuit* the sense behind the graph.
If you must only change the title, I admit that Matt’s “Cascading synonyms network” is accurate, if not exactly Plain Style.
I’m not sure I see the problem associating “dumb” with “black”. I don’t think the computer made an error in this case.
How about borrowing terminology from wordnet?? just call it sense network???? but of course, wordnet does a different thing anyways… but i still think it is a reasonable name…..
I got it – it is in fact an equivalence class of words which can be linked by synonym bridges, sometimes changing the intermediate words’ meanings, so why not call it an “equivocation class”?
An improvement can be “Word Association Network (synonym)”.
In this way, you can specify the kind of association in the network.
Dramatic chipmunk synonymous words of doom!?
Has a nice ring to it.
Synonyms: Teach W|A to use every curated bit of knowledge that comes its way not just parrot it.
You are teaching it all the synonyms. Assign each a seperate suffix to positively identify its meaning. When the user inputs a word with synonyms W|A is to return a dropdown list of the synonyms with their suffix attached and their meanings. Now ith a serious input containing many words the user can attach the W|A suffix where appropriate to indicate its precise meaning. The suffixes should be numerical selfchecking numbers to reduce keying mistrakes on entry.
Input: Dumb
W|A returns among the rest of its clarification of the input:-
Dumb(19) Dim meaning – dim
Dumb(27Silent meaning – silent
Dumb(35)Speechless mesning – speechless
Dumb(43) Unarticulate meaning – inarticulate
User not only selects the specific definition but may even remember it and next time enter dumb(27) or whatever.
Teach W|A the latest Grammer which hopefully W|A will improve and impose on Input..
That’s one enjoying post. Now let me check the synonym network for gizmodo.
I think the synonym network is the least useful part of the look-up-a-word page. It’s an interesting idea, just rarely useful, so it should appear somewhere other than the look-up-a-word page.
I’d love to be able to use W|A to play with graph structures, by the way 😉
An error in the last example input: 4th element, 102, should be dropped, 66 is the number of 2nd-level synonyms, 423 of 3rd etc.
On a side note, example would be more interesting in my opinion had you limited results to adjectives only. Then numbers go up as far as I calculated them correctly as 1, 8, 42, 176, 518 etc. with saturated value 4 402.
Which is surprisingly lower. (So much more nouns and other things then among your synonyms?) WordData (in my Mathematica 7.0) counts 21 538 different adjectives, which is another surprise, because then there must be other synonym networks apart from this one “dumb” can generate. I’ve found that starting with “zealous” gives a network with maximum elements of 33.
Which are those other networks?
Hint:
You should really check the synonym networks for adjectives only. Then start with itsy-bitsy, and you get a saturated network with 12 vertexes only, but all extremely powerful in wrong female hands/mouth.
I think Synonym networks should be renamed to avoid the impression that synonyms are transitive, which they are certainly not. Why don’t you use Wordnet synsets? Or are you?
Hi, we like very much Wolfram | Alpha, because the information is very complete.
Thank you very much 4 your work
Diseno Web Madrid
http://www.internetiza.net
Hm for the query ‘branding’ it only comes op with – ‘the act of stigmatizing’.
I feel that this is not the strongest function of WA…however, all the rest is just great!
hi,
It seems really promising..I just ask and i don’t have to go through link to see th answers..
Well done guys oyu are all genious !!
Now what about a real dictionary ?? Frnch to German for example ??
Synonym network is a pretty self-explanatory name. However, the confusion comes perhaps from the labels applied to the network.
IMHO, for the network to have a use other than generating nice pictures, only the direct synonyms should be labeled, so that no-one thinks that “dark” is a synonym of “acetify”, as I’m seeing right now. In addition, it should be made dynamical: by clicking on a synonym, it should become the center of the network, and its direct synonyms should be labeled.
So what you’re saying is that the ‘synonym’ relation isn’t transitive…
You should just call it ‘Six degrees of Kevin Bacon for words’ like you explained in your post…
As a frequent user of dictionaries & the like, the synonym network is disappointing to me. It seems to show labels for the original word and some number of nodes representing first level synonyms from zero upwards (as a pathological case, try “set”), then possibly second or higher level ones. As the text says, the information being shown is actually a tree. There are well-established ways of displaying trees, and the synonyn network used something else. If it’s conveying information that traditional trees don’t convey, I’m too dense to see what it is. The stuff about multi-level synonyms is interesting but as far as I can tell you need Mathematica to get it – I can’t get WA to give me the same information. So I don’t understand why that’s in a WA blog article!
Based on a few tries, the words listed in the synonyms pod are never used as labels in the synonym network. I don’t understand how the words chosen for use as labels are selected, and there are often sizable clumps with no label.
Sorry to bang on but it’s just a pretty picture at the moment. Try looking at the networks for raspberry, bum, fart, bump, node, network and flip – a range of very different pics, but apart from a different number of synonyms, what are the pictures telling us that we wouldn’t get from a trad tree diagram as used for folders and files in umpteen operating system GUIs?
not sure if i’m following, is that the ‘synonym’ relation not transitive?
Jacob