Thanks for the corrections - the first edition was done in the dark ages of publishing - believe it or not I had to send in hardcopy, so it had more than it’s share of problems, particularly in the code.

- Mark

Love the book. It’s been on my bookshelf since ‘92.

I was referring to it this morning, and thought I’d pass along some errata, in case you plan to come out with a revised edition sometime soon.

These are pertaining to the first edition.

Page 125, in your table, all 18 of your “greater than” signs should be “less than signs”. Example,
SPACE 1/10 0.00 >= r > 0.10
(no value of r can be both less than 0 and greater than 0.1)

Page 234, second last paragraph ends with:
So this token is encoded as: 14, 4, ‘ .’ - which should be
So this token is encoded as: 14, 4, ‘ ‘.
(I’m sure some English-expert editor (or program) incorrectly put the period inside the quote for you. If you want to satisfy the editor, perhaps put it inside of double quotes, like you did on page 235:
So this token is encoded as: “14, 4, ‘ ‘.”

I found a few more minor things… not sure if you want them. Very helpful book, though! Thanks!

Actually, I don't have anything in book about Context Trees or the Context Tree Algorithm - because I don't know anything about them. Sorry.

Post to comp.compression and ask for references, I'm sure the good people there will be happy to give you a pointer or two.

- Mark

now i'm doing my final exam about lossless compression using context tree modeling. i have download several references, but until now i'm still confuse about context modeling and context tree algorithm. can you give me suggestion, a book or an article about this algorithm? thank you

Note that the head of the first table says "Ratios" - the numbers you are seeing are compression ratios. The ratio used is described in the front of the book in the section "Keeping Score". A value of 100 indicates perfect compression, and 0 indicates no compression.

However, Appendix A (Compression Statistics) gives some numbers without describing their units. There are some 4 digit numbers of the form NN.NN. I assume they are compression ratio percentages, but perhaps they are execution time in seconds! If they are percentages, they might be compressed size / original size or (original size - compressed size) / original size. Also, there are numbers listed between 3 digits and 5 digits long in other tables, are these bytes/second, total seconds, or something else? Maybe someone that understands could respond or an entry could be made under Errata.

best regard