Teaching myself and making an algorithm to assist me in reading effectively. I have not done it much up to this point, so it's okay to fail / be slow!
=== Sublime Video === (no adjustable playback speed)
Looks good so far. However, how to add support for variable speed playback?
I really like Mortimer Adler's suggestions in How To Mark A Book. Underline major statements. Vertical lines at margins for details.
And some lines and arrows to connect the main ideas together in one big mind map.
Word embeddings, word2vec
Look up CoreNLP (the Stanford group)
Deep learning works really well for fuzzy definitions that are hard to pin down in normal ways.
Figure out what you want to do first!
n-grams allow you to capture the nearby context of the word. E. Coli would be one n-gram instead of separate words.
arxiv.org, he checks it weekly! Is there a way to get to know the *good papers* from there? Federated wiki might allow you to make meta-notes, instead of assuming commenting allowed on that website.
Uses word embedding space to do deep learning on. But it's only using 1 word, why not n-gram?
Read Yav Goldberg natural language processing paper, best overview paper in his opinion.
Hypothes.is seems like most similar project. Another is PeerLibrary.
Liquid Software guy based on Engelbart's work. His Author app more to come on his twitter in November 2014 has a lot of similar features, but it should be on the web! (not just OS X). Also, did an excellent Engelbart documentary.
Draft is another similar thing to this project, but not that similar.
Support for Table of Contents arguably not necessary as we will auto-generate it from the text
Like most of the other research tools out there, except hopefully easier/better.
Assist with de-constructing a book into its “skeleton”/outline format, assisted with your notes and observations. And…maybe even improved with other people's advice.
j/k = up/down = scroll up and down to next sentence h = left = zoom out to next biggest outline l = right = zoom in to think about idea more. Open area on right more?
e = del = archive sentence (unnecessary detail). make it gray at next to bottom zoom level. Hide it at higher levels.
Already read sentences dim slightly
Force summaries to check understanding?
GET RID OF SEARCHING FOR MENU OPTIONS. SEARCH FOR THE COMMAND, PRESS THE MOUSE, '?' for KEYBOARD HINTS,etc.
“Document Management Service”, NetDocuments.
Paste buffer in Firefox/Chrome should use Web Events paste and access Clipboard. Cross-browser example in Pasteboard (but only does images, and doesn't work in IE?).
IE and Chrome work well enough with test file in project. Firefox, something isn't firing right.
They choose to balance theory and practice.
After completing these materials, then students will be ready to attempt one of the more advanced textbooks, such as Speech and Language Processing by Jurafsky and Martin
|Claim||The moon is made of cheese||is/was|
Therefore, the moon is made of cheese.
Stuff about synonym tables, stemming, etc. Sounds cool, not sure what they mean quite yet
Incoporate OpenNLP for named entity recognition (people, places, companies). Probably also “he, it, she”, etc.
Lucene has “More like this” functionality
Also, “SpanQuery”, only return the part of document that is relevant
BoostingTermQuery changes weights of terms in query
Caching common searches and showing the “right document” as best result is a nice hack.
recognizes logic, reasoning, fallacies of logic, presents a contrasting view. “Semantic ” ?
It'd be nice to figure out what makes a good explanation and make sure that you read books with good explanations first.
TLDR: Find a person you can trust and email them / search their blog or wiki directly :)
Train using Amazon reviews and the stars given. More tips on edge cases by pulling from Google papers and patents. It would be helpful to search the Google search results for Brian Klemmer (and other life coaches like him) and understand dissenting opinions and facts.