One question I've never seen an answer to: Why is China attracting so much more foreign investment than India? To an outsider, they both seem to put the same premium on science and engineering, and they both seem to have large pools of cheaper labor. Yet China has several times the dollar amount of foreign direct investment.
Could it be because India is a democracy and at least partially looks after its rural communities and environment, whereas the autocratic Chinese government can promise businesses protection from labor unrest and environmental regulation? Or are the Indians too protectionist with regard to foreign ownership?
I wonder if it would be any easier to make a binary-to-binary translator on the same architecture? The idea would be to translate legacy i386/i586 binaries to take advantage of the latest CPU extensions.
Any complicated self modifying code could be left the same if the program could at least spot it reliably. It might even be possible to translate to 64-bit at the cost of a few "x &= 0xFFF..." instructions around shift operations.
It's getting increasingly difficult to fund online poker accounts, which are enormously popular in light of the World Poker Tour and other televised events.
It looks like were seeing a new era of regulation through threat of regulation. The offshore drug sites are providing a valuable service too: AIDS activists lobbied to be allowed to import personal-use supplies of experimental drugs not yet approved domestically. They're also the main source of nootropics like Piracetam and Hydergine.
Blog/post segmentation into entries. Organize into expandable tree form with browser-side filtering of slashdot-type posts based on a learned per-site model of segmentation boundaries.
learn to prioritize entries according to user preference
click-through probability or time-on-screen estimates
using bag-of-words n-gram model
selectively display of all funny>3 posts, for example
show all posts/threads mentioning openbsd
this would be especially useful on freerepublic
should be able to classify dom-nodes, don't need to segment text
Find messages semantically similar to the one I'm about to post
TF/IDF would probably do well enough to be useful
Segmentation of content from stylesheet/advertising boilerplate
Auto security zone changes based on perceived threat level of site
frequently visited sites can be allowed to use cookies/js
would benefit from pagerank-type aggregate measure
search-query aware auto-summarization of pages
right-click on a google link, and select "open in new tab and
summarize", will get an expande form of the google summary
with page keywords/phrases and matching text/dom nodes at the
top, and the original page below
auto-open and integrate the follow-on pages of a multi-page article, especially the current 15 page hardware reviews. Remove the duplicate stylesheet junk.
"don't show me things like this" auto-inference of dom nodes or regexes to rewrite the page to hide parts of the boilerplate and create a leaner page. alternately, "extract this region of the page and incorporate it into a composite page" to allow the user to build his own rss-type portal from several pages, which can be automatically refreshed and bookmarked
clustering of history pages by content or referrer - DMOZ or inferred LSI categories, and (independently) viewing the browse tree google->somepage->someotherpage->foo.jpg
other non-ml feature ideas:
detach and drag/drop tabs between windows or to their own window. also to reorder them.
search across all tabs, with "find next" moving to the tab containing the next match
fuzzy search matching by stemming, or near matches to catch misspellings
Slashdot Mirror
One question I've never seen an answer to: Why is China attracting so much more foreign investment than India? To an outsider, they both seem to put the same premium on science and engineering, and they both seem to have large pools of cheaper labor. Yet China has several times the dollar amount of foreign direct investment.
Could it be because India is a democracy and at least partially looks after its rural communities and environment, whereas the autocratic Chinese government can promise businesses protection from labor unrest and environmental regulation? Or are the Indians too protectionist with regard to foreign ownership?
I wonder if it would be any easier to make a binary-to-binary translator on the same architecture? The idea would be to translate legacy i386/i586 binaries to take advantage of the latest CPU extensions.
Any complicated self modifying code could be left the same if the program could at least spot it reliably. It might even be possible to translate to 64-bit at the cost of a few "x &= 0xFFF..." instructions around shift operations.
It's getting increasingly difficult to fund online poker accounts, which are enormously popular in light of the World Poker Tour and other televised events.
It looks like were seeing a new era of regulation through threat of regulation. The offshore drug sites are providing a valuable service too: AIDS activists lobbied to be allowed to import personal-use supplies of experimental drugs not yet approved domestically. They're also the main source of nootropics like Piracetam and Hydergine.
4. It's not based on a video game. Of course, Event Horizon meets all these criteria too.
Anyone comfortable with scripting languages should be able to use the Gnu R statistics package and the GGobi visualization package to get the same effect in a cross-platform, free-as-in-speech way.
What a great name for software. Does this remind anyone else of the "No Va" urban legend?
Blog/post segmentation into entries. Organize into expandable tree form
with browser-side filtering of slashdot-type posts based on a learned
per-site model of segmentation boundaries.
learn to prioritize entries according to user preference
click-through probability or time-on-screen estimates
using bag-of-words n-gram model
selectively display of all funny>3 posts, for example
show all posts/threads mentioning openbsd
this would be especially useful on freerepublic
should be able to classify dom-nodes, don't need to segment text
Find messages semantically similar to the one I'm about to post
TF/IDF would probably do well enough to be useful
Segmentation of content from stylesheet/advertising boilerplate
Auto security zone changes based on perceived threat level of site
frequently visited sites can be allowed to use cookies/js
would benefit from pagerank-type aggregate measure
search-query aware auto-summarization of pages
right-click on a google link, and select "open in new tab and
summarize", will get an expande form of the google summary
with page keywords/phrases and matching text/dom nodes at the
top, and the original page below
auto-open and integrate the follow-on pages of a multi-page article, especially the current 15 page hardware reviews. Remove the duplicate stylesheet junk.
"don't show me things like this" auto-inference of dom nodes or regexes to rewrite the page to hide parts of the boilerplate and create a leaner page.
alternately, "extract this region of the page and incorporate it into a composite page" to allow the user to build his own rss-type portal from several pages, which can be automatically refreshed and bookmarked
clustering of history pages by content or referrer - DMOZ or inferred
LSI categories, and (independently) viewing the browse tree google->somepage->someotherpage->foo.jpg
other non-ml feature ideas:
detach and drag/drop tabs between windows or to their own window. also to reorder them.
search across all tabs, with "find next" moving to the tab containing the next match
fuzzy search matching by stemming, or near matches to catch misspellings
trap crashes and dump the list of open tabs