Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Multi-Layer Dictionary (2016)

learnthesewordsfirst.com

Post reply on HN
Print view View original

91–100 of 114 posts

Re: Multi-Layer Dictionary (2016)

#91

Post by stared »

jpamata wrote: post #47
grenoire wrote: post #2

Has there been any attempts to 'translate' this into other languages? I'm struggling most often with vocabulary first, whereas the grammar is much easier for me to grasp (programming helps?)

It's nothing like it but as a beginner, among the best books I've read about vocabulary are the "Madrigal Magic key to X" series of books by Margarita Madrigal. In it, she breaks down how to convert English words into French, German, etc. If I can recall correctly, here are a few simple examples on French: 1) Words ending with -or: replace the -or with -eur professor –> professeur, aggressor –> aggreseur 2) Words end…

I was told that there are many English-French false friends.

Vide Casually Explained on French: https://www.youtube.com/watch?v=a69toGGjoO0

Re: Multi-Layer Dictionary (2016)

#92

Post by phonebucket »

tom_mellior wrote: post #83
phonebucket wrote: post #17

If true, this is interesting from an academic perspective: word meanings can be derived from a space of 60 dimensions. But I’m not still not convinced of the value with respect to language learning. Learning how language is spoken from the fundamental 60 words sounds like trying to learn mathematics from its fundamental axioms. It seems like you might just get caught in a long list of definitions where you might be f…

> a space of 60 dimensions I don't think that's true. How would a 60-dimensional vector capture all the syntactic relationships between the words in a complex sentence?

Yeah, I was sloppy and wrong with my wording. Thanks for pointing it out.

I meant a sequence of one-hot encoded vectors of length 60 could make a good model of English.

Re: Multi-Layer Dictionary (2016)

#93

Post by quietbritishjim »

mcbits wrote: post #75
spicerguy wrote: post #71

Earlier quoted context omitted.

British English speaker here - I'm having difficulty identifying the word you're referring to here. Or are you referring more to definition and statistical presence in common usage?

Since "gotten" isn't in the list, I'm going with "someplace". It even seems unusual to me as an American, as I would write either "some place" or "somewhere". If the phrase is also unusual in British English, that's definitely something I had never noticed.

Is "gotten" the only word that exists in American English but not British English (apart from "someplace")? Your comment strikes me as trying to bring up a pet hate in an unrelated conversation.

Re: Multi-Layer Dictionary (2016)

#94

Post by __MatrixMan__ »

Robin_Message wrote: post #55
ErotemeObelus wrote: post #21

Earlier quoted context omitted.

I am a newbie to topology, so please help me. What is a topological sort and why does it describe this?

Not sure why it's called topological sort or what it had to do with topology, but it's a graph traversal where predecessors are visited before their successors. Such a sort only exists if there are no cycles in the graph.

It's been a while since I took a Topology class, but I left it with the understanding that Topology is all about taking a set (nodes, say) and applying some notion of closeness (how many hops?).

Once you've done that, you have a topological space. The sets that you've defined as "close together" are your open sets. Loosely speaking, smaller open sets are considered closer together than larger ones.

This might seem boring in the case of directed graphs, which have an obvious notion of closeness, but there are more exotic spaces where you might not usually think to use "spacial" reasoning, but where it can be applied anyhow thanks to the formalism.

In a topological sort, you end up arranging the elements based on nearness to each other (as much as is possible for a list), so that elements in a given open set in the typical topological space over directed graphs are likely to be next to each other in the list. It's a pretty topological way to do things, which probably explains the name.

Re: Multi-Layer Dictionary (2016)

#95

Post by leecarraher »

i once solved a similar problem with set theoretic approach using matroid theory and the greedy method for constructing a matroid basis using a thesaurus as my independence oracle. This was over a decade ago, so I no longer recall the results, but it certainly seems similar in the goal of finding a primal set of words that can in some way define all others.

Re: Multi-Layer Dictionary (2016)

#96

Post by __MatrixMan__ »

ww520 wrote: post #65
ErotemeObelus wrote: post #21

Earlier quoted context omitted.

I am a newbie to topology, so please help me. What is a topological sort and why does it describe this?

The "topological" in topological sort is more related to "network topology" than the mathematical topology (open sets). "Sort" is related to ordering. Topological sort thus is related to the ordering of a network graph's nodes by their edges. You can see https://en.wikipedia.org/wiki/Topological_sorting to see what it actually is. Topological sort is good in dealing with dependency graph. It can turn a dependency gra…

I don't think mathematical topology and network topology are different concepts.

The open sets in a network topology are the subsets of nodes that happen to be connected by the edges. So if you can't get from here to there without using a node outside your set, then your set isn't an open set in the network topology.

Re: Multi-Layer Dictionary (2016)

#97

Post by dbmueller »

trox wrote: post #88
phonebucket wrote: post #17

If true, this is interesting from an academic perspective: word meanings can be derived from a space of 60 dimensions. But I’m not still not convinced of the value with respect to language learning. Learning how language is spoken from the fundamental 60 words sounds like trying to learn mathematics from its fundamental axioms. It seems like you might just get caught in a long list of definitions where you might be f…

> If true, this is interesting from an academic perspective: word meanings can be derived from a space of 60 dimensions. But I’m not still not convinced of the value with respect to language learning. More like 60 x N , since each of those words can appear arbitrary many times.

Monoid with 60 generators. But is it free?

Re: Multi-Layer Dictionary (2016)

#98

Post by dbmueller »

__MatrixMan__ wrote: post #96
ww520 wrote: post #65

Earlier quoted context omitted.

The "topological" in topological sort is more related to "network topology" than the mathematical topology (open sets). "Sort" is related to ordering. Topological sort thus is related to the ordering of a network graph's nodes by their edges. You can see https://en.wikipedia.org/wiki/Topological_sorting to see what it actually is. Topological sort is good in dealing with dependency graph. It can turn a dependency gra…

I don't think mathematical topology and network topology are different concepts. The open sets in a network topology are the subsets of nodes that happen to be connected by the edges. So if you can't get from here to there without using a node outside your set, then your set isn't an open set in the network topology.

Except unions of open sets are required to be open, so the case for topology here is not what you're describing (assuming standard definitions).

Generally, the topology on a graph will look like what you get if you draw your graph in the euclidean plane (or space, whatever) without intersections of edges.

The notion of closeness of vertices in this case isn't really well described by "point set topology", but you'd rather use the notion of distance between vertices (length of a shortest path). But even then, the distance stuff generally assumes non-directed edges, because you generally want the distance from A to B to be the same as from B to A.

In short, I don't think "point set topology" has much to say here, at least as usually done.

Re: Multi-Layer Dictionary (2016)

#99

Post by yorwba »

goldenkey wrote: post #86
yorwba wrote: post #76

Earlier quoted context omitted.

How do you determine whether your magnetic field is pointing up and not down?

Closer and farther can be used to explain up and down.

How do you use closer and farther to determine whether a magnetic field is pointing up or down? It's not like there are little arrows on magnetic field lines you can just look at.

Re: Multi-Layer Dictionary (2016)

#100

Post by abecedarius »

crazygringo wrote: post #90

It's conceptually interesting, but it also strikes me as a problem that doesn't need solving. Having worked with and taught foreign language, nobody learns the first 1,000 words of a language from a same-language dictionary, nor should they. Children learn from the world; adults learn from classes or a translating dictionary. (Only intermediate/advanced level learners start to use a native dictionary.) The idea of "b…

I didn't start from zero, but this children's dictionary of French in French was useful to me learning it: https://www.amazon.com/Mon-premier-dictionnaire-Roger-Pillet...

All the words it uses are defined within it -- of course with some circularity, but it's heavy on examples and pictures. It was intended for nonnative children taking classes in a style more like native immersion than is typical in schools. I wish more resources followed this philosophy.

Post reply on HN