long tail terms had high competition but very low search volume
Remember what the competition column is: it measures the number of AdWords clients bidding on those terms. As an advertiser, I typically do not want to waste my ad money on short-tail terms. I want to capture the searcher who is already way down the sales funnel.
Let's say I'm a local Ford dealer. I'm not looking to help build Ford's brand. I'm looking for the guy who sees the sun is out, spring is coming and he just absolutely wants a new car and he already knows more or less which one he wants.
Am I currently bidding on - cars (no) - brand (definitely not) - model (definitely not) - "brand model" (definitely not) - "year brand model submodel" (probably not unless there's a national market because this particular car is hard to find and sold out in many locales) - "year brand model submodel" city
So there might not be much competition in the AdWords tool for car or Ford, but when you get down to a specific model in a specific city, the bid price is likely to go way up, because these are buying terms.
But from an SEO perspective, will it be harder to rank #1 for "cars" or "year brand model submodel" city
cars - 11,640,000,000 results "year brand model submodel" city - 14,400 results
So those might be competitive terms because they are money terms, but from an SEO perspective, their much easier to rank for.
The problem is that if I go to Google Insights or the Keyword Tool, I see there is not enough worldwide search data for Google to even report on those terms I'm playing with.
Then you have other questions to look at
1. It still might be hard if there are not that many pages, but they're all powerhouses like wikipedia as in your example (or in my example, like Ford.com and Car and Driver or what have you).
2. How specific are the terms? A lot of single-word phrases have multiple meanings. Back to my example, many car models have other meanings too. If someone enters [mustang] in Google she might be thinking of buying a horse or looking for employment at a famous ranch in Nevada.
3. How optimized are all those pages? You can very quickly get some idea of how many of the pages are minimally optimized for those terms by comparing
"year brand model submodel" with intitle:"year brand model submodel"
3. You can look to go long tail on short tail terms if you have some images, videos, etc that rank you in universal search.
Back to my example, I get 601K for "year brand model submodel" but only 1450 on image search and, typically, image search will slide in around number three or four in universal search and it will show the top four images if they're all landscape format, the top six if they're all portrait, and the top 5 if there's a mix.
So to be in the top four in Google for a non-image, I actually need to be in the top three (because one of the top four is usually for images, though perhaps not for your terms, but charts and graphs and infographics rank well in image search too).
This means I need to be in the top 0.0005% to show in the top four of universal search with a text-page match. To be in the top 4 of the universal search results by ranking with an image, I only need to be in the top 0.3% of images. Granted, there are downsides to images (no clickable title; video has them though!), but it can be a way to squeeze into shorter tail terms with a longer tail strategy.
That's a bit rambly, but just trying to throw out some ideas for how you can attack the glossary problem and see if you can't start ranking for some of those terms.