Morris Ernst was an ACLU attorney who challenged other high-profile book bans during the period. He found a cooperative publisher and hatched the plan to import a copy of the book from France and have it seized by customs, although the getting it seized was more difficult than anticipated.
> A few days later the book showed up at Random House—it had passed through customs. Furious, Ernst personally marched the package over to the customs office and demanded that it be searched. When the inspector opened it and found Ulysses, he muttered, “Oh, for God’s sake, everybody brings that in. We don’t pay attention to it.” Ernst insisted that he seize it. On May 8, the book was officially seized by customs.
In lieu of the usual fees, he would receive 5% of the book royalties should it be legalized and published (not a bad deal).
> 1. Use only the beginning of the document, as that's probably the most important part anyways, and it's fast.
That seems to be a solution devised for news articles, as the standard news writing style involves providing answers to the Five Ws up front on the article.
I'll add to this that you can add a very crude (separate) model for the document length and number of distinct words, and use that to flag outlier documents that might bump into the known weaknesses with respect to document length.
This is not invariant to the size of the document (though agreed, generally better). It doesn't solve the problem of having mostly positive features and a negative prior.
Stated more formally, your model is b + wᵀx. Generally, b is < 0, and E[wᵀx] > 0. As the document grows, wᵀx tends to dominate b. You'll have bias with length as long as E[wᵀx]≠0 and there aren't any constraints on w that would force this.
If your data obeys the naive Bayes assumptions then this model is mathematically optimal. That each word is independently drawn from some distribution conditional on it's class. E.g. if there was an exactly 1% chance any given word in a spam email would be "viagra".
Now obviously real world data doesn't obey these assumptions perfectly. But I don't see how violating the independent features assumption would cause the problem you mention. A longer email does mean the word "viagra" is more likely to occur in a normal email just by random chance. But the model takes that into account by recording the frequency of "viagra" in normal emails and seeing if it's consistent with that.
For a simple example, imagine a dataset where the naive assumption is true if you split it into 100 classes, but false if you split it into one vs everything else. All of the conditional probabilities for the "everything else" class will be underestimated, biasing the weights towards the one.
This problem happens because the class you are interested in is more compact than its inverse.
It's also exacerbated by feature selection, as the negative features have smaller weights and thus lower information gain than the positive features.
I'm in the same situation, my office has an open floor plan and can get very distracting at times. +1 for 8tracks, I haven't created any myself but there are some great electronic playlists with minimal vocals for coding.
I think that the social aspect of it is actually what makes it fun here. Quality playlist is a hard thing to find and editors' picks are not always the most suitable ones
