If someone asked me what I perceive to be my greatest personal shortcoming it would be my poor understanding and facility with mathematics, which usurps only slightly my monolingualism.
I suck really badly at math and it, more than anything, has been the biggest stumbling block in my professional career and personal study. Virtually every field of endeavour that seems “obvious” given my personal inclinations is more or less sealed shut by my crappiness at math.
For example, in the process of trying to tackle a problem that should be VERY straightforward to someone with a strong mathematical ability and knowledge of self organizing maps. Except, I know next to nothing about self-organizing maps and all the literature I’ve come across is totally opaque, filled as it is with what appear to be highly organized insecta squashed flat between the pages.
I have grown so far behind that appears I might ALWAYS be this bad at it.
A simple geometrical transformation problem still bamboozles me (anyone with a solution, please comment with an explanation). If I take a square with corners at points A, B, C and D and I know the length of a side and the coordinates for point A on a cartesian plane, and I rotate this square 90 degrees clockwise, what are the new coordinates of A? What about the coordinates for any given angle? What about for an arbitrary (rather than perfectly centered) pivot point?
This is something I swear I learned in middle-school but it’s just not there.
The current problem I’m grappling seems simple to solve. I’m trying to determine the quantifiable relationship between specificity and relevancy. This requires prior knowledge of a semantic taxonomy - where some tokens in a string will have a descendent/predecessor relationship - a tree of specificity, going towards the trunk you get more general and going to the leaves you get more and more specific.
But how is that related to relevancy? An easy assumption is that the trunk, the most general, is 0% relevant to everything and any leaf is 100% relevant to matching content. So it simply 100% / the number of links between leaf and tree?
In the case of text documents. If a web page is ALL about Toyota Rally Raid Cars it’s 100% relevant to “Toyota Rally Raid Cars” but how relevant is it to “Cars” or “Toyota”? How does the relevancy scale?
If we’re looking at tokens that means the document is 25% relevant to each of the tokens “Toyota” “Rally” “Raid” and “Cars”. Is this meaningful?
How do I solve this? Why can’t I be better at math?